Hydrogen Generation and Coke Formation over a Diesel Oxidation

Dec 8, 2010 - A Study on Remanufacturing of Deactivated Commercial Diesel Oxidation Catalyst by CVS-75 mode in Light Duty Diesel Engine. Chang-Hee ...
0 downloads 0 Views 1MB Size
Download PDF
1156

J. Phys. Chem. C 2011, 115, 1156–1162

Hydrogen Generation and Coke Formation over a Diesel Oxidation Catalyst under Fuel Rich Conditions† Meshari AL-Harbi,‡ Jin-Yong Luo,‡ Robert Hayes,§ Martin Votsmeier,| and William S. Epling*,‡ Department of Chemical Engineering, UniVersity of Waterloo, Waterloo, ON N2L 3G1 Canada, Department of Chemical and Materials Engineering, UniVersity of Alberta, Edmonton, Alberta, Canada, T6G 2G6, and AutomotiVe Catalysis DiVision, Research and DeVelopment, Umicore, Hanau, Germany ReceiVed: July 13, 2010; ReVised Manuscript ReceiVed: NoVember 15, 2010

Hydrogen production via hydrocarbon steam reforming and water gas shift reactions was investigated over a monolith-supported Pt-based diesel oxidation catalyst. The evaluation included comparison between constantly rich gas composition conditions and cycling between rich gas conditions and an inert stream. Analysis was performed along the catalyst length at temperatures ranging from 200 to 500 °C. During the constant inlet composition experiments, C3H6 steam reforming started at 375 °C, while dodecane steam reforming began at 450 °C, and resulted in less hydrogen produced. With a mixture of C3H6 and dodecane, hydrogen production originated solely from C3H6 steam reforming and, under otherwise identical conditions, was less than that observed with only C3H6, but higher than that with only dodecane. Hydrogen production from the water gas shift reaction was higher than that observed with hydrocarbon steam reforming and started at 225 °C. During cycling experiments, hydrogen production via hydrocarbon steam reforming was higher than that observed during the constant inlet composition experiments. This improvement was observed at all temperatures. Temperature programmed oxidation experiments performed after steam reforming indicate coke formed on the catalyst surface during steam reforming, and that the coke deposits were primarily toward the upstream portion of the catalyst. The data also show that the reason for better performance during cyclic operation is that less coke was deposited compared to that during noncyclic experiments. 1. Introduction Diesel oxidation catalysts (DOC) are used in a variety of leanburn engine aftertreatment systems. They are typically installed upstream of selective catalytic reduction (SCR) and NOX storage/ reduction (NSR) catalysts. Their function in such systems is to oxidize engine-out NO to NO2, as NO2 is trapped more readily than NO on NSR catalysts1,2 and a 1:1 NO:NO2 ratio promotes the “fast” SCR reaction over SCR catalysts.3 DOC are also installed upstream of diesel particulate filters, again to oxidize NO to NO2. NO2 is more reactive toward soot than O2, lowering the soot oxidation temperature by approximately 200 °C.4 Literature evidence shows that the temperature range in which DOC operate overlaps the temperature range in which steam reforming reactions are possible. Both water and hydrocarbons are present in diesel exhaust, providing reforming reactant species. And since DOC contain Pt and Pd supported on either alumina or zeolites, steam reforming reactions are likely, especially during the reductant-rich phase of a NSR cycle. Hydrogen can be produced via numerous catalytic methods, including catalytic partial oxidation, autothermal reforming, or steam reforming of hydrocarbons, alcohols, and biomass.5-9 Hydrocarbon steam reforming is typically the preferred process for industrial-scale hydrogen production, because it does not require oxygen, operates at relatively low temperature, and maintains a higher product H2/CO ratio than that of autothermal reforming and catalytic partial oxidation.10 The CO produced †

Part of the “Alfons Baiker Festschrift”. University of Waterloo. § University of Alberta. | Umicore. ‡

during steam reforming can also be used in the water gas shift (WGS) reaction to drive the production of extra hydrogen. Several studies have shown that the metal type11-13 as well as the support type can influence the extent of steam reforming. Alumina is a typical support in catalysts but tends to induce coke formation during steam reforming due to its surface acidity,14 with coke formation a primary steam reforming deactivation process.15,16 However, the presence of precious metals, zirconium, or other alkaline components in the catalyst formulation can minimize coke formation17 or facilitate its removal.18 Hydrocarbon steam reforming over Pd and Pt has been extensively investigated. Previous results relevant to the current study include C3H8 steam reforming over Pd/CeO2/Al2O3 and Pt-Rh/CeO2/Al2O3 catalysts,19,20 with the reaction starting at about 350 °C. Steam reforming of C3H6 and 2-propanol was also investigated over a powder Pd-Cu/γ-Al2O3 catalyst,21 with both reactions starting at 327 °C and increasing steadily until complete conversion was attained, which under the conditions of the tests, was 527 °C. For NSR catalysts, hydrogen has repeatedly been reported better than other reductant species (CO and hydrocarbons) in reducing surface NOX species to N2.22-26 Therefore, if the amount of hydrogen can be increased via steam reforming or WGS in the upstream DOC during the regeneration phase, which is reductantrich relative to oxygen, the NOX conversion to N2 over the downstream NSR catalyst could be improved. The main interest in this study is comparing and quantifying the amount of hydrogen formed during cyclic and noncyclic (constant steam reforming or WGS conditions) operation over a diesel oxidation catalyst. Coke formation and regeneration via coke removal was also investigated after these experiments to explain the observed differences.

10.1021/jp106472p  2011 American Chemical Society Published on Web 12/08/2010

Hydrogen Generation and Coke Formation

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1157

2. Experimental Methods In this study, a commercial monolith diesel oxidation catalyst supplied by Umicore AG was used. The sample contains 95 g/ft3 Pt supported on Al2O3. The monolith block that the sample was removed from had a cell density of 400 cpsi. The sample was 2.3 cm in diameter with a length of 6 cm. The sample was wrapped in 3M insulation material and placed into a horizontal quartz tube reactor, which was in turn placed inside a Lindberg Minimite temperature-controlled furnace. The matting was used to seal the gap between the catalyst and reactor wall. Two K-type thermocouples were placed at the radial centers of the catalyst; one at the inlet face and one at outlet edge of the catalyst. A third was placed ∼2.5 cm upstream of the sample. During the experiments, the monolith remained nearly isothermal, with only ∼2 °C temperature differences observed between the inlet and outlet face. All gases except balance N2 were supplied by Praxair. The N2 was produced using an On-Site nitrogen generator system. Bronkhorst mass flow controllers were used to meter gases to the reactor system. The dry gas mixture was then heated to >120 °C, and water was then introduced using a Bronkhorst CEM system. In experiments that included dodecane or m-xylene, they were also metered with a Bronkhorst CEM system and introduced downstream of the water injection location, closer to the reactor, thereby eliminating any reactions with the steel tubing walls. Small quartz tubes, 3 mm o.d. and 2 mm i.d., were placed in the front portion of the furnace and before the catalyst to help in heat transfer and limit fully developed flow. In the constant gas composition steam reforming and WGS experiments, 5% H2O, 0.27% hydrocarbon on a C1 basis or 0.27% CO, and a balance of N2 were used. In cycling experiments, 60 s inert and 10 s reductant-containing phases were used (labeled rich below, matching that of an NSR cycle). In the inert phase, 5% H2O and a balance of N2 were used while in the rich phase, 5% H2O, 0.27% CO or hydrocarbon on a C1 basis, and a balance of N2 were used. In cycling experiments, the rich and inert gas mixtures were made in separate manifolds. A fourway, fast-acting solenoid valve was used to switch between the two. Experiments were also performed to investigate coke formation during cycling and noncycling experiments. C3H6 steam reforming experiments were performed between 300 and 500 °C, which typically took about 3 h in total to complete, or at one temperature depending on the experiment. Upon completion of a steam reforming experiment, the reactor was cooled to 50 °C in N2 and then 10% O2 was added to the feed and the reactor was ramped to 500 °C at rate of 7 °C/min. Catalyst regeneration from deposited coke by O2, H2O, H2, and a mixture of H2O and H2 after C3H6 steam reforming experiments were evaluated, using the same TPO protocol, but substituting in the other regeneration species for the O2. Experiments were performed with a 25 000 h-1 space velocity at standard conditions. The gas compositions were measured using a MKS MultiGas 2030 FTIR analyzer. Spatially resolved capillary-inlet mass spectrometry (SpaciMS) was also used to measure H2, H2O, and hydrocarbons along a radially centered monolith channel. In these studies, He was added and used as a tracer for calibration purposes. To resolve the gas concentrations spatially, a silica capillary, connected to the sampling end of a capillary from a Hiden Analytical mass spectrometer, was placed within one of the radially centered catalyst channels. The capillary dimensions were 0.3 mm i.d. and 0.43 mm o.d. Gases were collected at different positions by moving the silica capillary tip to different positions within the channel.

Figure 1. H2 concentrations obtained at different temperatures and lengths of the catalyst during steam reforming experiments. The inlet gas composition was 900 ppm C3H6, 5% H2O, and balance N2.

3. Results and Discussion 3.1. H2 Generation during Noncyclic Conditions. Hydrogen generation via steam reforming was investigated between 300 and 500 °C. In the steam reforming experiments, C3H6, C12H26, m-C8H10, and mixtures of these were selected to represent different hydrocarbon species in diesel exhaust. In this series of experiments, 5% H2O, 0.27% hydrocarbon on a C1 basis, and a balance of N2 were used. Also in these experiments, spatially resolved capillary-inlet mass spectrometry (SpaciMS) was used to quantify the amounts of the species and study their axial concentration distribution. In m-C8H10 steam reforming experiments, no hydrogen was detected in the outlet stream at temperatures as high as 500 °C and therefore data associated with m-C8H10 are not shown. Figure 1 shows the amount of hydrogen formed at different catalyst positions during the C3H6 steam reforming experiments. Note, the position labeled zero is just inside the inlet face of the catalyst (really about 1 mm). C3H6 steam reforming began at 375 °C, though to a very small extent. As the temperature was increased stepwise to 500 °C, hydrogen formation monotonically increased. It is also clear from Figure 1 that hydrogen production increased as a function of catalyst length. In a recent study,27 C3H6 steam reforming was investigated over a model Pt/BaO/Al2O3 NSR catalyst. Hydrogen formation started at a similar temperature (375 °C), but to a higher extent than observed in this study due to the presence of Ba. Alkaline materials are known to suppress the acidity of the alumina support28 and thus to reduce coke formation, which leads to higher hydrogen formation. Hydrogen generation via dodecane steam reforming was also investigated and the outlet hydrogen concentration data are shown in Figure 2. Below 450 °C, no steam reforming occurred and beyond that hydrogen production was observed, starting at a higher temperature than that observed with C3H6, and increased at 475 and 500 °C. Also, the amount of hydrogen formed at 500 °C with dodecane was 10, 100, and 209 ppm at 0, 3, and 6 cm from the catalyst face, respectively, whereas the amount of H2 formed with C3H6 at the same temperatures and locations were 15, 210, and 488 ppm, or about 2 times more by the outlet with C3H6. These results demonstrate that the extent of steam reforming over a diesel oxidation catalyst depends on the type of hydrocarbon used. In previous studies,13,29-33 alkane steam reforming was investigated over Pt, Pd, Rh, Ru, and Pt/Rh catalysts. According to these studies, alkane steam reforming occurs at higher temperatures than those observed with alkenes,

1158

J. Phys. Chem. C, Vol. 115, No. 4, 2011

Figure 2. Outlet H2 concentrations obtained at different temperatures and with different hydrocarbon feed mixtures during steam reforming experiments. The inlet gas composition was 900 ppm C3H6, 225 ppm C12H26, or 900 ppm C3H6 and 225 ppm C12H26, 5% H2O, and balance N 2.

consistent with our data, although different hydrocarbon chain lengths were used in the present study. Similar experiments were also carried out with a mixture of C3H6 and dodecane. The quantified outlet hydrogen amounts are shown in Figure 2. During these experiments, no change in the dodecane amount was observed, indicating that dodecane steam reforming did not occur. The C3H6 amount monotonically decreased as a function of temperature and catalyst length (data not shown); consequently, the hydrogen measured during these experiments originated exclusively from C3H6 steam reforming. It should be noted that hydrogen production started at 400 °C, slightly higher than that observed with only C3H6 (375 °C). This decreased hydrogen production in the presence vs absence of dodecane was observed at all temperatures tested. For example, at 500 °C and in the absence of dodecane, the outlet hydrogen measured was 488 ppm while in the presence of dodecane it was 285 ppm. These results demonstrate that there was mutual inhibition between C3H6 and dodecane, resulting in lower hydrogen production. Maillet et al.34 investigated hydrocarbon steam reforming over Rh, Pt, and Pd supported on Al2O3 catalysts. The authors divided the steam reforming process into three main steps. The first step involves dissociative adsorption of the hydrocarbon on the metal sites, the second includes dissociative adsorption of water on the support, and finally, OH groups from the support migrate to the metal particles to react with a CHx fragment originating from the dissociative adsorption to yield CO2 and H2. This mechanism explains the mutual inhibition between dodecane and propylene observed. C3H6 adsorbs strongly on the Pt sites, relative to dodecane, possibly dissociating to a CHx fragment and blocking access of dodecane to the Pt. Adsorption on the support is less selective, and some of the dodecane adsorbs on the catalyst support, which in turn inhibits water dissociation, or inhibits migration of OH groups to the metal site, ultimately inhibiting the reaction between OH groups and the CHx fragment. TPD of adsorbed dodecane (data not shown) further supports that dodecane is adsorbed on the catalyst surface, as it was observed desorbing up to ∼400 °C. Hydrogen production via the water gas shift (WGS) reaction was also investigated, between 200 and 500 °C. In this series of experiments, 5% H2O, 2700 ppm CO, and a balance of N2 were used. The results are shown in Figure 3, along with equilibrium H2 concentrations for the conditions tested. It is clear that the WGS reaction started at a lower temperature (∼225 °C) than steam

AL-Harbi et al.

Figure 3. Outlet H2 concentrations obtained at different temperatures during noncycling and cycling water gas shift reactions experiments. The inlet gas composition was 2700 ppm CO, 5% H2O, and balance N2.

reforming for the hydrocarbons tested. SpaciMS data show that hydrogen concentrations increased as a function of catalyst length and temperature, although the rate of increase slowed at higher temperature with the approach to equilibrium. The amount of hydrogen formed by the outler via the WGS reaction at 500 °C was 3.5 times higher than that observed with C3H6 steam reforming and 6 times higher than that observed with dodecane steam reforming. In the WGS reaction, for every mole of CO reacted, 1 mol of hydrogen will form and therefore at 500 °C approximately 60% of the CO was consumed in the formation of hydrogen. No methanation was observed, and CO conversion was only associated with H2 and CO2 production. This finding is in contrast to previous studies,35,36 where methane formation started at temperatures >375 °C with a Pt/CeO2 catalyst during WGS experiments. However, methanation depends on numerous factors such as inlet gas composition, metal loading, and the inclusion of promoters such as alkali and cerium oxide components. In the previous studies,35,36 significantly larger amounts of both CO and H2O were used in the feed mixtures (e.g., 11.4% CO and 45.7% H2O) coincident with including H2, while in this study CO and H2O concentrations were 0.27% and 10%, respectively, with no H2 added. Furthermore, in the present study the catalyst does not contain cerium oxides or other promoters while in the previous studies, the catalyst included cerium oxide, and as stated by the authors, this would provide extra adsorption sites for water rather than CO blocking all surface sites. Therefore, using an excess amount of CO and H2O with a catalyst containing cerium oxide would lead to high C/H ratios at the catalyst surface, driving the reactions toward methanation. 3.2. H2 Generation during Cycling Conditions. Hydrogen production via the steam reforming and WGS reactions during cycling experiments was also investigated at different temperatures. Steam reforming is typically carried out under steady-state inlet conditions; however, for diesel aftertreatment NSR applications, the feed is cycled between those of normal engine exhaust and those of the rich phase, where significant steam reforming is possible. In the “inert” phase, which lasted 60 s, 5% H2O and a balance of N2 were used, while in the rich phase, which lasted 10 s, 5% H2O, 0.27% CO or hydrocarbon on a C1 basis, and a balance of N2 were used. An inert phase rather than a true lean phase was used to better understand the phenomena occurring along the catalyst during the rich phase. In both steam reforming and WGS experiments, SPACiMS was used to quantify gas-phase concentrations and study their axial distribution at three different catalyst positions. In the plotted data, the front position is at the inlet of the sample (∼1 mm in). Propylene steam reforming during cycling was performed between 300 and 500 °C. The measured amounts of hydrogen

Hydrogen Generation and Coke Formation

Figure 4. H2 concentrations obtained at different temperatures and lengths of the catalyst during cycling steam reforming experiments. The inert phase gas composition was 5% H2O and balance N2. The rich phase gas composition was 900 ppm C3H6, 5% H2O, and balance N2. The inert phase was 60 s and the rich phase was 10 s. The front position represents the inlet of sample (∼1 mm in).

Figure 5. Outlet H2 concentrations obtained at different temperatures and with different hydrocarbon feed mixtures during cycling steam reforming experiments. The inert phase gas composition was 5% H2O and balance N2. The rich phase gas composition was 900 ppm C3H6, 225 ppm C12H26, or 900 ppm C3H6 and 225 ppm C12H26, 5% H2O, and balance N2. The inert phase was 60 s and the rich phase was 10 s.

produced are shown in Figure 4. C3H6 steam reforming again started at 375 °C, the same as for the noncyclic experiments. The hydrogen formed increased as a function of catalyst length and temperature. Two primary differences were observed in these experiments when compared with noncyclic C3H6 steam reforming. The first is that a significantly higher amount of hydrogen formed at the front of the catalyst during cycling compared with noncyclic experiments, where the maximum amounts of hydrogen with all temperatures tested did not exceed 15 ppm. For example, at 500 °C, the hydrogen production at the front of catalyst during cycling experiment was ∼10 times higher than that observed under steady inlet conditions. Another observation is that the amounts of hydrogen formed at the middle and outlet positions in the catalyst during cylcing were also higher, but to a lesser extent, being about 2 times higher than that observed during steady-state experiments. These differences are likely due to more coke being deposited along the catalyst during noncyclic tests, especially at the upstream portion, which will be discussed in the following section. Similar experiments were carried out, but with dodecane, and the H2 generation results shown in Figure 5. Again, dodecane steam reforming started at 450 °C and the hydrogen amount progressively increased with catalyst length. Although there were differences in the amounts of hydrogen formed during cycling and noncycling

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1159 experiments, they were not as significant as those observed with C3H6. For example, at 500 °C, the hydrogen formed in the middle and outlet of the catalyst was 100 and 210 ppm during noncycling experiments, while 130 and 300 ppm were formed during cycling experiments. Again, however, more significant differences were observed at the front of the catalyst, where again the hydrogen amount was higher with cycling experiments. The smaller differences between dodecane cycling and noncycling steam reforming experiments compared with those observed with C3H6 is related to the hydrocarbons type, where C3H6 is activated faster and more easily than dodecane, and therefore steam reforming and associated coke formation occur at lower temperatures. The positive effect of cycling is less evident for dodecane due to the higher temperatures required for the onset of reaction, and at these higher temperatures, as will be shown below, coke can be reacted from the surface. Hydrogen production with a mixture of C3H6 and dodecane was also investigated and the quantified amounts of hydrogen are shown in Figure 5. The conditions are otherwise similar to those experiments with either C3H6 or dodecane. Again, no change in the dodecane amount was observed while the C3H6 concentration steadily decreased along the length and with increasing temperature. This indicates that the hydrogen formed was again solely due to C3H6 steam reforming. The hydrogen amounts formed were also again lower than those observed in the absence of dodecane. The reason as stated earlier is because of the mutual inhibition between C3H6 and dodecane. It should also be pointed out that hydrogen production was observed at 375 °C, which is 25 °C lower than those observed during noncyclic experiments (Figure 2). This suggests that dodecane inhibition was mitigated by the cycling conditions, likely by desorption from the surface during the inert phase. Additionally, the amount of hydrogen formed during cycling experiments was about twice the amount formed during noncycling experiments. The lower coke buildup on the catalyst during the 10 s rich phase of the cycle followed by the inert phase, as will be shown in the next section, explains the higher hydrogen production and lower temperature during cycling. The extent of hydrogen formed via the WGS reaction during cycling was also investigated at different temperatures and different catalyst positions. The hydrogen formed during the 10 s rich phase was measured and the results are shown in Figure 3. Again, hydrogen production started at 225 °C and the hydrogen amount increased as a function of catalyst position and temperature. In previous studies,37,38 the WGS reaction was investigated during cyclic operation over a commercial NSR catalyst. The extent of the WGS reaction was 10% at 200 °C and 81% at 500 °C, which is higher than those observed in this study. The higher WGS reaction extent in the previous studies is due to the presence of excess O2 (10%) in the lean phase, thereby removing any residual CO adsorbed during the previous rich phase, and the presence of Ce and alkali and/or alkaline earth elements (e.g., Ba), which are known to enhance the WGS reaction39-41 and suppress coke formation.31,32 During cycling experiments with the DOC in this study, the trends were relatively similar to those observed during the noncycling WGS experiments, except at the front of catalyst, where a higher amount of hydrogen was observed with cycling. Compared with hydrocarbon steam reforming during cycling, the hydrogen formed during cycling for the WGS reaction was still higher along all catalyst positions and all temperatures. Takahashi et a.l42 studied C3H6 steam reforming and WGS reactions over a NSR catalyst between 200 and 400 °C under cycling conditions. They also showed that the amount of H2 formed during the WGS reaction was higher than that formed during C3H6 steam reforming,

1160

J. Phys. Chem. C, Vol. 115, No. 4, 2011

Figure 6. Outlet H2 concentrations obtained during C3H6 steam reforming experiments at 450 °C. The inlet gas composition was 900 ppm C3H6, 5% H2O, and balance N2.

consistent with our data. But, it should be mentioned that the amount of H2 formed via the C3H6 steam reforming and WGS reactions was higher than those observed in our experiments and attained at lower temperatures. In Takahashi’s study, 7% O2 was used in the lean phase and the catalyst contained alkali and alkaline earth elements (Ba and K) and ceria-zirconia-based oxygen storage components. Therefore, during the rich pulse, the stored or residual oxygen would lead to combustion of some CO and C3H6, possibly resulting in a temperature increase within the catalyst bed. Thus, the measured temperatures for hydrogen production in the previous study would be lower than those observed in our study and hence might explain the observed differences. Another contributing factor is that the O2 in the lean phase removes coke deposited during the previous rich phase, or adsorbed CO, which should not affect the reaction onset temperature but will affect the amounts of H2 or conversions observed. 3.3. Coke Formation during Steady-State and Cycling Experiments. As discussed above, there were no significant differences in outlet hydrogen formed for the WGS reaction between cycling and noncycling conditions. However, hydrogen generation was higher during steam reforming under cycling conditions compared to that observed under constant inlet feed steam reforming conditions. To determine if coke formation is the reason for this latter difference, coke formation during steam reforming experiments was investigated. Ethylene was observed during catalyst outlet measurements, and since ethylene is considered a coke precursor during steam reforming, coke formation is likely. And although coke formation is a product of complete hydrocarbon decomposition, CO is the standard reforming product, and at the reforming temperatures, CO is not considered a catalyst poison, while coke deposition does lead to deactivation because it reduces the effective surface area.43 It is likely that the amount of coke deposited on the catalyst during noncyclic steam reforming experiments was ultimately higher than that during the cycling steam reforming experiments. Further evidence includes the time-resolved H2 concentration profiles, with an example shown in Figure 6, where the H2 measured by mass spectrometry dropped over time during the noncyclic C3H6 steam reforming experiment. C3H6 steam reforming was selected for these tests. In the first set of experiments, noncyclic C3H6 steam reforming experiments were performed between 300 and 500 °C under conditions identical to those in Figure 1. Upon completion of the steam reforming experiments at 500 °C, the reactor was cooled to 50 °C in N2 and then 10% O2 was added to the inlet gas feed and the reactor was ramped to 500 at 7

AL-Harbi et al.

Figure 7. CO2 formation obtained during temperature programmed oxidation experiments at 2 and 4 cm from the front of the catalyst. After C3H6 steam reforming experiments, the reactor was cooled to 50 °C with only N2 and then 10% O2 was added to feed and the reactor was ramped to 500 °C at rate of 7 °C/min.

°C/min. The capillary for SpaciMS readings was placed at two locations; 2 and 4 cm from the inlet face, to investigate the amount of coke deposited during the test (two experiments were performed, with measurements taken at 2 cm during the first and 4 cm during the second). The amount of evolved CO2 during these TPO experiments was measured and the results are shown in Figure 7; no CO was observed. The amounts of C deposited on the catalyst were quantified on the basis of the CO2 evolved and were 394 and 395 µmol upstream of 2 and 4 cm, respectively. As shown, oxygen begins to remove the coke from the catalyst as low as 190 °C. Coke removal reached a maximum at 230 °C and coke was completely removed by 250 °C. Such data suggest that during normal NSR cycling, less inhibition would be observed, at least above 230 °C, from coke buildup because O2 will be available during the lean phase to remove any built-up coke from the previous rich phase of the cycle. The data shown in Figure 7 also show that the amount of formed CO2 is quite similar at the 2 and 4 cm positions, indicating that the deposited coke was only in the front 2 cm of the catalyst. On the basis of the data shown in Figure 1, the reaction is still occurring downstream and is a combination of steam reforming and WGS, with the WGS originating from CO formed during steam reforming. The constantly higher hydrocarbon concentrations at the front of the catalyst lead to larger coke deposits there, and a drop in formation down the length as the hydrocarbon is consumed. Furthermore, hydrogen is known to suppress coke formation during steam reforming,44 and thus the product hydrogen leads to decreasing amounts of coke observed down the length of catalyst. With the test ending at 500 °C, these effects are even more pronounced due to the higher reaction rates and therefore more reaction at the inlet. To isolate the effect of the temperature and further investigate coke deposition along the catalyst length, noncyclic C3H6 steam reforming experiments were performed at 450 °C and for different reaction times; 4, 60, and 180 min. The reactor was subsequently cooled to 50 °C in N2, 10% O2 was then added, and the reactor was ramped to 500 at 7 °C/min. The capillary for SpaciMS was placed at 4 cm from the catalyst front. As expected, coke formation increased as the reaction time increased. These results indicate that coke deposition builds as a function of reaction time, at least to 180 min, which is consistent with the data shown in Figure 6. The formed CO2 during TPO was measured to quantify the amount of C on the catalyst surface. The amounts of C deposited on catalyst surface

Hydrogen Generation and Coke Formation

J. Phys. Chem. C, Vol. 115, No. 4, 2011 1161

Figure 8. CO2 formation obtained during temperature programmed oxidation experiments at 1, 2, and 4 cm from the front of catalyst. After cycling C3H6 steam reforming experiments at 375 °C, the reactor was cooled to 50 °C with only N2 and then 10% O2 was added to the feed and the reactor was ramped to 500 °C at rate of 7 °C/min.

Figure 9. CO2 formation data obtained during temperature programmed reduction experiments at 4 cm from the front of catalyst. After noncyclic C3H6 steam reforming experiments during a temperature programmed ramp from 300 to 490 at 1 °C/min, the reactor was cooled to 50 °C with only N2 and then (A) 5% H2O was added to the feed, or (B) 1000 ppm H2 and 5% H2O were added to feed, and the reactor was ramped to 525 °C at a rate of 7 °C/min.

were 269, 422, and 887 µmoles at 4, 60, and 180 min, respectively. CO2 measurements during TPO after noncyclic C3H6 steam reforming experiments performed at 450 °C for 180 min were obtained at two different locations: 2 and 4 cm. The amounts of C deposited on catalyst surface were 354 and 887 µmol at 2 and 4 cm, respectively. The amount of coke at 4 cm was almost 2.5 times that at 2 cm. Due to the integral nature of the monolith, coke will be deposited first at the front of the catalyst and further verifies that coke deposition was spread along the catalyst length with this longer reaction time. Similar experiments were also performed after the cycling C3H6 steam reforming experiments, run at 375 °C and under conditions otherwise identical to those described for Figure 4. During cycling experiments, no CO2 was observed evolving during the inert phase, up to ∼500 °C. After the cycling experiment, the reactor was cooled to 50 °C and then ramped at 7 °C/min to 500 °C with 10% O2 and balance N2. The capillary was placed at three different locations (thus steam reforming was repeated at 375 °C three times): 1, 2, and 4 cm from the catalyst front. The detected CO2 during TPO was measured, and the results are shown in Figure 8. At 1 cm, the maximum amount of CO2 measured was about 12 ppm whereas 19 and 20 ppm were detected at 2 and 4 cm, and the amounts of C deposited on the catalyst surface were 31, 91, and 102 µmol at 1, 2, and 4 cm, respectively. Again, these data demonstrate that coke deposition occurred more at the front portion with small reaction times, here over the front 2 cm of the catalyst during the 10 s of C3H6 steam reforming. Coke deposition was also primarily observed in the front 2 cm with longer times in the noncyclic runs mentioned above, but in those experiments the temperature was higher, resulting in increased rates at the front as well. For the sake of a more direct comparison, cycling and noncycling C3H6 steam reforming experiments were performed at 400 °C. The noncyclic experiment was held for 50 min, and three 10 s rich cycles for the cycling experiment were performed. This admittedly leads to significantly less exposure during the cycling experiments, but on the basis of the TPO data presented above, any built-up coke could be oxidized at these temperatures in any case. The reactor was cooled to 50 °C and then ramped at 7 °C/min to 500 °C with 10% O2 and balance N2. The capillary was placed 4 cm from the catalyst front. The C amount was measured during TPO after 50 min was 385 µmol, while 137 µmol was measured after the 10 s cycling experiment. It is apparent that significantly

less total carbon was deposited during the briefer cycling experiments, although definitely nonlinear with respect to time, which ultimately leads to the increased H2 production observed. 3.4. Regenerating the Catalyst from Deposited Coke. Numerous regeneration methods have been proposed in the literature to remove coke deposited on catalysts. Oxygen, H2O, CO2, and H2 are the most commonly used gases for removing coke.44 In this study, catalyst regeneration from deposited coke by O2, H2O, H2, and a mixture of H2O and H2 was investigated after noncyclic C3H6 steam reforming experiments. Although the TPO data show coke removal at relatively low temperatures, it is necessary to determine if the other gas components play a critical role in coke removal as well, to understand and model such phenomena. The steam reforming conditions prior to the regeneration tests are similar to those described in Figure 1. In all of these experiments, the mass spectrometer capillary was placed at 4 cm from the catalyst front. In one experiment, C3H6 steam reforming was performed during a temperature programmed ramp from 300 to 490 °C at a rate of 1 °C/min. Subsequently, the reactor was cooled in N2 to 50 °C. The reactor was then ramped up at a rate of 7 °C/min to 525 °C with a feed containing 5% H2O and the balance N2. The CO2 measured during the ramp is shown in Figure 9. Water began to remove coke at 495 °C, and the catalyst was completely “cleaned” at 525 °C. Similar experiments were also performed, but the regeneration mixture during the temperature ramp portion of the experiment contained 1000 ppm H2 and 5% H2O (a H2 value similar to those observed during steam reforming experiments). No CH4 or CO was detected in the product stream. The CO2 was observed at the same temperature when regenerating with only H2O, as shown in Figure 9. This indicates the 1000 ppm of H2 had little, if any, impact on regeneration although it suppresses coke formation.44 Further experiments were also carried out to investigate the ability of just H2 to remove coke. C3H6 steam reforming was carried out at 400 °C, to eliminate the effect of H2O on coke removal, since H2O begins to react with the coke at ∼495 °C, as was shown in Figure 9. C3H6 was turned off after 50 min of steam reforming and then 500 ppm of H2 was introduced for 20 min. If the H2 was able to remove the deposited coke, CH4 should be detected in the outlet stream via the following reaction: C + 2H2 f CH4. However, no CH4 was observed in the outlet stream. One possible reason is that the formed CH4 could be re-formed in the presence of H2O to H2 and CO or CO2 via the following reactions:

1162

J. Phys. Chem. C, Vol. 115, No. 4, 2011

CH4 + H2O f CO + 3H2 CH4 + 2H2O f CO2 + 4H2 But, neither CO nor CO2 was observed in the outlet stream, indicating that CH4 steam reforming also did not occur. To verify that the coke was not removed from the catalyst, the reactor was cooled to 50 °C in N2, 10% O2 was then added, and the reactor was ramped to 500 at 7 °C/min. CO2 was detected in the outlet stream (data not shown), peaking at 75 ppm at about 225 °C, indicating that the H2 did not clean the catalyst from deposited coke in the presence of H2O. As further evidence for the lack of H2 reaction with surface coke, a TPR experiment was performed after C3H6 steam reforming during a temperature ramp from 300 to 480 °C. The reactor was cooled afterward in N2 to 50 °C and then ramped at 7 °C/min to 525 °C with a feed containing 1000 ppm H2 and a balance of N2. No CH4, CO, or CO2 was detected in the outlet stream, again indicating that H2 did not remove the deposited coke from the catalyst. The reactor was subsequently cooled in N2 to 50 °C, and then 10% O2 was introduced and the reactor was ramped to 500 at 7 °C/min. Again, CO2 was detected in the outlet stream. Overall, these data demonstrate that H2 does not react with deposited coke on this DOC at temperatures as high as 525 °C. H2 could be able to regenerate the catalyst at T > 525 °C, but this was not investigated to avoid catalyst aging. Furthermore, these data support previous conclusions regarding the effect of H2 in suppressing coke formation, such that as H2 is produced via steam reforming, less coke forms, consistent with the observed axial gradient in surface C along the length of catalyst. 4. Conclusions Hydrocarbon steam reforming and water gas shift reactions were investigated as a function of catalyst length and temperature over a monolith supported diesel oxidation catalyst. Hydrogen production was measured during both cycling and noncycling (constant steam reforming and water gas shift reaction conditions) experiments using spatially resolved capillary-inlet mass spectrometry (SpaciMS). The data demonstrate that hydrogen production with C3H6 steam reforming started at 375 °C, while dodecane steam reforming occurred at 450 °C and with less hydrogen produced. When C3H6 and dodecane were present together, mutual inhibition was observed and the hydrogen formed only originated from C3H6 steam reforming. The amount of hydrogen formed via the WGS reaction was higher and started at a lower temperature (∼225 °C) than that observed with hydrocarbon steam reforming. The amount of hydrogen formed during cycling hydrocarbon steam reforming experiments was consistently higher than that obtained from the noncycling experiments. Coke deposition was investigated during both types of experiments and the results show that higher amounts of coke were deposited during the noncycling experiments, compared to cycling experiments, providing the reason for the observed differences in hydrogen formed. Coke deposition was found to start at the front of the catalyst and spread downstream as the reaction time increased. Acknowledgment. We thank the Natural Sciences and Engineering Research Council of Canada Discovery Grant

AL-Harbi et al. Program, Auto21, and Kuwait University for financial support, and Umicore for the sample provided. References and Notes (1) Epling, W. S.; Campbell, L. E.; Yezerets, A.; Currier, N. W.; Parks, J. E. Catal. ReV. 2004, 46, 163. (2) AL-Harbi, M.; Epling, W. S. Catal. Today 2009, 130, 121. (3) Grossale, A.; Nova, I.; Tronconi, E.; Chatterjee, D.; Weibel, M. J. Catal. 2008, 256, 312. (4) Setiabudi, S. A.; Makkee, M.; Moulijn, J. A. Appl. Catal. B 2004, 50, 185. (5) Wei, J.; Iglesia, E. J. Phys. Chem. B 2004, 188, 4094. (6) Navarro, R. M.; Alvarez-Galvan, M. C.; Sanchez-Sanchez, M. C.; Rosa, F.; Fierro, J. L. G. Appl. Catal. B 2004, 55, 223. (7) Navarro, R. M.; Penˇa, M. A.; Fierro, J. L. G. Chem. ReV. 2007, 107, 3952. (8) de Lima, S. M.; da Cruz, I. O.; Jacobs, G.; Davis, B. H.; Mattos, L. V.; Noronha, F. B. J. Catal. 2008, 257, 356. (9) Abbas, H. F.; Daud, W. M. A. W. Int. J. Hydrogen Energy 2010, 35, 1160. (10) Holladay, J. D.; Hu, J.; King, D. L.; Wang, Y. Catal. Today 2009, 139, 244. (11) Rostrup-Nielsen, J. R. J. Catal. 1973, 31, 173. (12) Rostrup-Nielsen, J. R.; Alstrup, I. Catal. Today 1999, 53, 311. (13) Kolb, G.; Zapf, R.; Hessel, V.; Lo¨we, H. Appl. Catal. A 2004, 277, 155. (14) Sanchez-Sanchez, M. C.; Yerga, R. M. N.; Kondarides, D. I.; Verykios, X. E.; Fierro, J. L. G. J. Phys. Chem. A 2010, 114, 3873. (15) Bartholomew, C. H. Appl. Catal. A 2001, 212, 17. ¨ nsan, Z. Catal. ReV. 2001, 43, 31. (16) Trimm, D. L.; IlsenO (17) Ghenciu, A. Curr. Opin. Solid State Mater. Sci. 2002, 6, 395. (18) Montini, T.; De Rogatis, L.; Gombac, V.; Fornasiero, P.; Graziani, M. Appl. Catal. B 2007, 71, 125. (19) Faria, W.; Dieguez, L. C.; Schmal, M. Appl. Catal. B 2008, 85, 77. (20) Barbier, J.; Duprez, D. Appl. Catal. A 1992, 85, 89. (21) Resini, C.; Arrighi, L.; Delgado, M.; Vargas, M.; Alemany, L. J.; Riani, P.; Berardinelli, S.; Marazza, R.; Busca, G. Int. J. Hydrogen Energy 2006, 31, 13. (22) Jozsa, P.; Jobson, E.; Larsson, M. Top. Catal. 2004, 30-31, 177. (23) James, D.; Fourre′, E.; Ishii, M.; Bowker, M. Appl. Catal. B 2003, 45, 147. (24) Poulston, S.; Rajaram, R. R. Catal. Today 2003, 81, 603. (25) Lesage, T.; Verrier, C.; Bazin, P.; Saussey, J.; Daturi, M. Phys. Chem. Chem. Phys. 2003, 5, 4435. (26) Liu, Z.; Anderson, J. A. J. Catal. 2004, 224, 18. (27) AL-Harbi, M.; Epling, W. S.; Radtke, D. Appl. Catal. B 2010, 96, 524. (28) Bolis, V.; Magnacca, G.; Morterra, C. Res. Chem. Intermed. 1999, 25, 25. (29) Igarashi, A.; Ohtaka, T.; Motoki, S. Catal. Lett. 1991, 13, 189. (30) Lia, Y.; Wanga, X.; Xiea, C.; Songa, C. Appl. Catal. A 2009, 357, 213. (31) Wang, S.; Lu, G. Q. Appl. Catal. B 1998, 19, 267. (32) Monteiro, R. S.; Dieguez, L. C.; Schmal, M. Catal. Today 2001, 65, 77. (33) Yao, Y.-F. Y. J. Catal. 1984, 87, 152. (34) Maillet, T.; Barbier, J.; Duprez, D. Appl. Catal. B 1996, 9, 251. (35) Wheeler, C.; Jhalani, A.; Klein, E. J.; Tummala, S.; Schmidt, L. D. J. Catal. 2004, 223, 191. (36) AL-Harbi, M. Epling, W. S. Yezerets, A. Currier, N. W. Chen, H.-Y. Hess, H. SAE Technical Paper Series 2009-01-0631. (37) AL-Harbi, M.; Epling, W. S. Appl. Catal. B 2009, 89, 315. (38) Venezia, A. M.; Pantaleo, G.; Longo, A.; Carlo, G. D. J. Phys. Chem. B 2005, 109, 2821. (39) Letichevsky, S.; Tellez, C. A.; Avillez, R. R.; da Silva, M. I. P.; Fraga, M. A.; Appel, L. G. Appl. Catal. B 2005, 58, 203. (40) Fu, Q.; Weber, A.; Flytzani-Stephanopoulos, M. Catal. Lett. 2001, 77, 87. (41) Takahashi, N.; Yamazaki, K.; Sobukawa, H.; Shinjoh, H. Appl. Catal. B 2007, 70, 198. (42) Adhikari, S.; Fernando, S.; Haryanto, A. Energy Fuels 2007, 21, 2306. (43) Kern, C.; Jess, A. Chem. Eng. Sci. 2005, 60, 4249. (44) Forzatti, P.; Lietti, L. Catal. Today 1999, 52, 165.

JP106472P